How we go faster
If you want to get into technical detail about how we move data faster, then you're in the right spot. If you want to know more about our business value, check out our homepage:
A CDN is all about the infrastructure
We took a look around at how other CDNs, both public and private, were being built and operated. We asked what we could feasibly do better, how we could workaround problems that other companies ran into.
The goal: make data as small as possible and move it as quickly as possible, served from as close as possible to the end users... with little to no processing latency.
Enter: Caddy, the ultimate web server.
Caddy is an open-source web server that is the powerhouse behind our CDN service.
You should use Caddy, it's awesome.
Whether or not you use our service, you should look into how you might be able to implement Caddy for your app. It's convenience, stability, and performance are unmatched.
Learn more about Caddy on their site:
We leverage Caddy in multiple ways to run our CDN service, from automatically provisioning and distributing SSL certificates, to compressing and serving all of our edge traffic.
Caddy takes care of our edge traffic - that's where things needs to be as fast as possible since these servers are closest to the user.
We use Redis as a distributed cache layer for web data. The Redis database is distributed between every node in the same datacenter.
Nginx takes care of the backend aggregation traffic - this is where data flows into our network from your origin servers. We use Nginx to cache, minify, inject code, and more.
What about your hardware?
This one was all effort: We put in the time to look for every single datacenter service provider we could find, and grew a sizable list of information about their offerings.
Tons of service providers were evaluated and once we decided what the initial network would look like, we began to build it out to prove the concept. We evaluate more providers everyday so we know what locations we are ready to grow into.
As of today, Skip2 has found countless providers and has evaluated well over 100. You might be curious about the hardware we run, which we describe below:
For AMD, we require a minimum of a Zen-3 based (2020-released) CPU. For Intel, we require a minimum of Skylake Bronze and Silver CPU's (2017-released). For RAM, of course, compatible DDR4 with Error Correct Code (ECC) technology to detect and correct data errors.
We scale each node up or down over time.
All of this illustrates how our low-overhead architecture enables us to use lower-power equipment to serve data fast.